Biography
Dr. Shuvodeep De is a Lecturer and Research Scientist in the Ingram School of Engineering at Texas State University, working where computational mechanics, artificial intelligence, and high-performance computing meet. His work is driven by a single question: how do we make physics-based simulation fast and trustworthy enough to design the materials, structures, and manufacturing processes of the future, not just analyze them?
Dr. De earned his Ph.D. in computational mechanics at Virginia Tech under Prof. Rakesh K. Kapania, developing multidisciplinary design optimization methods for aircraft wings with curvilinear spars and ribs. During his research appointment at Oak Ridge National Laboratory, he built simulation and optimization frameworks for large-scale metal additive manufacturing. At the University of Alabama, he modeled advanced electroplating processes in research sponsored by BASF, and his industry work with NASA (through M4 Engineering) and Metalsa produced lighter aircraft wings and truck chassis through computational design.
Today, he leads research on AI-accelerated multiscale mechanics, digital twins for additive manufacturing, and closed-loop design of advanced materials, in collaboration with research groups at Rice University and Texas State University. He has more than 20 peer-reviewed publications, an h-index of 13, and is co-inventor on U.S. Patent 12,661,466. At Texas State University, he teaches hands-on courses in mechanical behavior of materials and CAD and digital manufacturing, advises senior design teams, and helps lead the department’s ABET assessment.
Dr. De earned his Ph.D. in computational mechanics at Virginia Tech under Prof. Rakesh K. Kapania, developing multidisciplinary design optimization methods for aircraft wings with curvilinear spars and ribs. During his research appointment at Oak Ridge National Laboratory, he built simulation and optimization frameworks for large-scale metal additive manufacturing. At the University of Alabama, he modeled advanced electroplating processes in research sponsored by BASF, and his industry work with NASA (through M4 Engineering) and Metalsa produced lighter aircraft wings and truck chassis through computational design.
Today, he leads research on AI-accelerated multiscale mechanics, digital twins for additive manufacturing, and closed-loop design of advanced materials, in collaboration with research groups at Rice University and Texas State University. He has more than 20 peer-reviewed publications, an h-index of 13, and is co-inventor on U.S. Patent 12,661,466. At Texas State University, he teaches hands-on courses in mechanical behavior of materials and CAD and digital manufacturing, advises senior design teams, and helps lead the department’s ABET assessment.
Research Interests
Breaking the computational bottleneck in engineering. Simulation has become the primary way engineers design materials and structures, and it is reaching its limits. The most important problems ahead, such as materials that must survive hypersonic impact, 3D-printed parts whose residual stress must stay within tolerance, and architected materials engineered for a target response, require models that span length scales, couple several physics, and must be solved thousands of times inside optimization and uncertainty quantification. Their cost is growing faster than computers are improving. Since the end of classical processor scaling, raw speed no longer rescues slow algorithms: today’s exascale machines deliver most of their power through GPUs that traditional engineering codes cannot use well, and real-time digital twins demand answers in seconds from simulations that take days. Unless the algorithms change, simulation will become the bottleneck of engineering innovation rather than its engine.
My research tackles this problem directly by co-designing three things: physics-based models, machine-learning methods that avoid unnecessary computation, and algorithms built for modern hardware. The toolkit includes neural operators and surrogate models, Bayesian active learning, reduced-order models, adjoint and differentiable solvers, matrix-free and mixed-precision finite elements on GPUs, and AI agents that automate simulation workflows. One principle governs all of it: every fast model is verified against high-fidelity simulation or experiment before it is trusted.
This engine currently drives three research thrusts. Learning-accelerated multiscale and fracture mechanics bridges microstructure and component scale, from shock-induced transformation in diamond–cBN–cobalt composites to fracture of double-network elastomers. Real-time digital twins for metal additive manufacturing aims to predict and control residual stress and distortion fast enough to adjust a build as it happens. Closed-loop design and discovery of materials couples AI-driven search with 3D printing and testing, from battery materials to auxetic lattices and gecko-inspired dry adhesives for robotic grippers.
My research tackles this problem directly by co-designing three things: physics-based models, machine-learning methods that avoid unnecessary computation, and algorithms built for modern hardware. The toolkit includes neural operators and surrogate models, Bayesian active learning, reduced-order models, adjoint and differentiable solvers, matrix-free and mixed-precision finite elements on GPUs, and AI agents that automate simulation workflows. One principle governs all of it: every fast model is verified against high-fidelity simulation or experiment before it is trusted.
This engine currently drives three research thrusts. Learning-accelerated multiscale and fracture mechanics bridges microstructure and component scale, from shock-induced transformation in diamond–cBN–cobalt composites to fracture of double-network elastomers. Real-time digital twins for metal additive manufacturing aims to predict and control residual stress and distortion fast enough to adjust a build as it happens. Closed-loop design and discovery of materials couples AI-driven search with 3D printing and testing, from battery materials to auxetic lattices and gecko-inspired dry adhesives for robotic grippers.
Teaching Interests
I teach mechanics as a loop: derive the answer, compute it, and test it against a measurement. Students who master all three develop the engineering judgment that matters most in an era when AI makes computation cheap but verification essential. My courses combine hand analysis, finite element simulation, and hands-on experimentation, and I teach students to use AI tools transparently and with verification.
My teaching interests include Statics, Mechanics of Materials, Machine Design, Mechanical Vibrations, the Finite Element Method, computer-aided design and digital manufacturing, mechanics and materials laboratories, and capstone design. I am also developing graduate courses in Scientific Machine Learning for Mechanics and Multiscale Modeling of Materials.
My teaching interests include Statics, Mechanics of Materials, Machine Design, Mechanical Vibrations, the Finite Element Method, computer-aided design and digital manufacturing, mechanics and materials laboratories, and capstone design. I am also developing graduate courses in Scientific Machine Learning for Mechanics and Multiscale Modeling of Materials.